Commit History

18 days ago

  1. v3: fixture consistency fixes from the 2026-11-01 blind run (21 cases), SKILL.md warning-marker legend, README benchmark result (glm-5.3-flash vs nemotron-ultra-550b:free)

    Verified signature
    9422da5

19 days ago

  1. v2: resumable runner, blind-judge protocol hardening, round-5 dataset anonymization

    Verified signature
    20177bc
  2. v2: resumable runner, blind-judge protocol hardening, round-5 dataset anonymization

    Verified signature
    f45b781
  3. Initial public release: 120-case blind A/B evaluation kit

    Verified signature
    ff736b7
  4. initial commit

    Verified signature
    138db10